Enhanced Low-Temperature CO Oxidation on a Stepped Platinum

Henry D. Lewis, Daniel J. Burnett, Aaron M. Gabelnick, Daniel A. Fischer, and John L. Gland*. Departments of Chemical Engineering and Chemistry, Unive...
0 downloads 0 Views 230KB Size
J. Phys. Chem. B 2005, 109, 21847-21857

21847

Enhanced Low-Temperature CO Oxidation on a Stepped Platinum Surface for Oxygen Pressures above 10-5 Torr Henry D. Lewis,† Daniel J. Burnett,† Aaron M. Gabelnick,‡ Daniel A. Fischer,§ and John L. Gland*,†,‡ Departments of Chemical Engineering and Chemistry, UniVersity of Michigan, Ann Arbor Michigan 48109, and National Institute of Standards and Technology, Gaithersburg, Maryland 20899 ReceiVed: March 25, 2004; In Final Form: July 2, 2005

The rate of CO oxidation has been characterized on the stepped Pt(411) surface for oxygen pressures up to 0.002 Torr, over the 100-1000 K temperature range. CO oxidation was characterized using both temperatureprogrammed reaction spectroscopy (TPRS) and in situ soft X-ray fluorescence yield near-edge spectroscopy (FYNES). New understanding of the important role surface defects play in accelerating CO oxidation for oxygen pressure above 10-5 Torr is presented in this paper for the first time. For saturated monolayers of CO, the oxidation rate increases and the activation energy decreases significantly for oxygen pressures above 10-5 Torr. This enhanced CO oxidation rate is caused by a change in the rate-limiting step to a surface reaction limited process above 10-5 Torr oxygen from a CO desorption limited process at lower oxygen pressure. For example, in oxygen pressures above 0.002 Torr, CO2 formation begins at 275 K even for the CO saturated monolayer, which is well below the 350 K onset temperature for CO desorption. Isothermal kinetic measurements in flowing oxygen for this stepped surface indicate that activation energies and preexponential factors depend strongly on oxygen pressure, a factor that has not previously been considered critical for CO oxidation on platinum. As oxygen pressure is increased from 10-6 to 0.002 Torr, the oxidation activation energies for the saturated CO monolayer decrease from 24.1 to 13.5 kcal/mol for reaction over the 0.95-0.90 ML CO coverage range. This dramatic decrease in activation energy is associated with a simple increase in oxygen pressure from 10-5 to 10-3 Torr. Activation energies as low as 7.8 kcal/mol were observed for oxidation of an initially saturated CO layer reacting over the 0.4-0.25 ML coverage range in oxygen pressure of 0.002 Torr. These dramatic changes in reaction mechanism with oxygen pressure for stepped surfaces are consistent with mechanistic models involving transient low activation energy dissociation sites for oxygen associated with step sites. Taken together these experimental results clearly indicate that surface defects play a key role in increasing the sensitivity of CO oxidation to oxygen pressure.

1. Introduction The catalytic oxidation of carbon monoxide (CO) on platinum has been widely studied due to its technological importance and the apparent “simplicity” of this bimolecular surface reaction. 1,2 Structure-activity relationships for surface reactions have often been characterized by comparing reaction on a series of singlecrystal surfaces with well-established surface structures.3-5 Model single crystals with step and kink defects have played an important role in establishing the key role played by defects in surface reactions.5-7 Step sites generally adsorb molecules and atoms more strongly and often accelerate dissociation.8-14 In this paper we present new evidence that bimolecular oxidation of CO is accelerated on the stepped Pt(411) surface for elevated oxygen pressures. On platinum surfaces, the CO oxidation reaction proceeds via a bimolecular reaction mechanism.1,2 High CO surface coverages are thought to inhibit dissociative oxygen adsorption and under these conditions CO2 formation is limited by CO desorption, which controls oxygen adsorption rates. In excess * Address correspondence to this author. Department of Chemical Engineering, University of Michigan. Phone: (734) 764-7354. Fax: (734) 647-4865. E-mail: [email protected]. † Department of Chemical Engineering, University of Michigan. ‡ Department of Chemistry, University of Michigan. § National Institute of Standards and Technology.

oxygen and above the CO desorption temperature, dissociative oxygen adsorption is generally not inhibited and the reaction is surface reaction rate limited. The activation energy for CO oxidation generally decreases with decreasing CO surface coverage.15-18 Adsorbed CO is more mobile than adsorbed atomic oxygen, and surface diffusion of CO to active sites containing oxygen is generally considered the dominant reaction mechanism when both reactants are adsorbed. Carbon monoxide oxidation on platinum catalysts is generally regarded as a structure-insensitive reaction.1,2,19 However, several reports indicate that the oxidation rate of CO depends on the size of metal particles for supported platinum catalysts.20,21 On platinum single-crystal surfaces, the surface reaction clearly depends on the structure of the surface.6,15,17,22-24 For example, a single CO2 peak is formed between 320 and 340 K on extended low-index platinum surfaces over a large coverage range.15,17 On stepped single-crystal surfaces, multiple CO2 formation peaks ranging from 100 to 500 K have been observed.24-28 The multiple reaction pathways for CO2 formation on stepped surfaces provide a clear illustration of one way that defects add to the complexity of surface reactions. The oxidation of CO on surfaces with defect sites has been the focus of several UHV studies.24,25,27-37 The active site for CO2 formation on stepped surfaces depends on the surface

10.1021/jp0486696 CCC: $30.25 © 2005 American Chemical Society Published on Web 10/26/2005

21848 J. Phys. Chem. B, Vol. 109, No. 46, 2005 coverages of CO and oxygen.28,35 CO adsorbed on terraces is more easily oxidized than CO adsorbed at steps, and preferential formation of CO2 at terrace sites has been observed on stepped platinum surfaces.22,25,28,30,33,36 Formation of CO2 at step sites has been observed less frequently.28,30,33,34 Using infrared spectroscopy, oxygen atoms adsorbed at defect sites were found to preferentially react with coadsorbed CO on an imperfect Pt(111) surface.18 Step troughs were determined to be the most active sites for the electrochemical oxidation of CO on stepped platinum surfaces.38 Oxygen islands formed at defect sites were proposed to be active in the oxidation of a saturated CO coverage on the Pt(100) surface.39,40 Thus, surface defects create new reaction pathways even under UHV conditions. Many practical catalytic reactions are performed at or above atmospheric pressure. To bridge the pressure gap between fundamental mechanistic studies at low pressure and catalytic reactions at higher pressures, there has been considerable interest in elevated pressure mechanistic studies. Initially, ex situ elevated pressure methods using vacuum-based analysis methods were used.41 More recently, a number of in situ techniques have been developed so that the surface can be characterized while the reaction is proceeding at elevated pressures.5,42,43 Several in situ methods have been used to study CO on model surfaces with defects.44-47 This paper describes the application of powerful soft X-ray methods to characterize the oxidation of adsorbed CO in flowing oxygen. The in situ kinetic studies reported here for the CO oxidation on the stepped Pt(411) surface over broad temperature (1001000 K) and pressure (10-8 to 0.002 Torr) ranges clearly indicate the important role played by defects and substantial sensitivity to oxygen pressure has been observed for the first time. For oxygen pressures below 10-5 Torr, CO must desorb from terrace sites before CO oxidation is initiated at steps. For oxygen pressures above 10-5 Torr, both step and terrace CO can be oxidized at low temperature and the reaction becomes surface reaction limited even for the CO saturated surface. The acceleration of low-temperature CO oxidation on a stepped surface under elevated pressure conditions presented in this paper highlights the importance of characterizing surface reactions over a realistic range of surface conditions. 2. Experimental Section Temperature-programmed reaction spectroscopy (TPRS) and fluorescence yield soft X-ray experiments were combined to characterize the reaction of CO and oxygen on Pt(411) over an extended temperature and pressure range. TPRS experiments were conducted at the University of Michigan in a conventional ultrahigh vacuum (UHV) chamber and soft X-ray experiments were completed on the U7A experimental endstation at the National Synchrotron Light Source (NSLS), Brookhaven National Laboratory. The Pt(411) single crystal was obtained from Monocrystals Company (Richmond Heights, OH) as a circular disk (1 cm diameter × 2 mm thick). The surface structure of the Pt(411) surface is shown in Figure 1. The Pt(411) surface consists of alternate 2- and 3-atom wide (100) terraces separated by one-atom high (111) steps.48 The UHV chamber at the University of Michigan is capable of reaching a base pressure of 6 × 10-11 Torr. The chamber is equipped with a computer-controlled UTI 100C quadrupole mass spectrometer (QMS), low-energy electron-diffraction-Auger electron spectroscopy (LEED/AES) optics, an ion gun for surface cleaning, and an X-ray photoemission spectrometer (XPS). The Pt(411) sample was mounted on a resistive heating element attached to a liquid nitrogen cooled manipulator capable

Lewis et al.

Figure 1. Structure of the Pt(411) surface. The surface is constructed with 2- and 3-atom wide (100) terraces separated by (111) monatomically high steps.

of x-y-z and 360° rotational motion. The Pt(411) single crystal was attached by spot welding two 0.5 mm tantalum wires around the top and bottom edges of the sample. The crystal temperature was measured with a chromel-alumel type K thermocouple spot welded to the back center of the sample. The crystal could be cooled to 95 K and heated to 1050 K. The sample was cleaned by repeated Ar+ sputtering and by oxidative cycles in ∼5 × 10-8 Torr oxygen using temperatures ranging from 600 to 700 K followed by annealing to 1000 K for several minutes in a vacuum. Surface cleanliness was verified by AES and oxygen desorption experiments. Oxygen and CO were dosed using separate dosers for both the flowing oxygen and coadsorbed TPRS experiments. The CO used was obtained from Matheson Tri-gas (research grade) and the oxygen from Air Products (research grade). CO exposures were performed at 100 K using a 6.25 mm stainless steel doser positioned within 5 cm of the surface. For coadsorbed TPRS experiments, oxygen was directionally dosed with the sample 3 cm from the doser. The sample was then positioned within 2 mm of the mass spectrometer for heating. For flowing oxygen TPRS experiments the CO-preexposed sample was positioned within 5 mm of the mass spectrometer inlet. In this configuration the oxygen doser was located in the plane of the mass spectrometer and approximately 6 cm away from the sample. Oxygen flow was initiated and upon reaching the indicated oxygen pressure, the sample was heated linearly. A heating rate of 5 K/s was utilized for all TPRS experiments. Oxygen desorption could not be monitored for flowing oxygen experiments due to the high oxygen background signal. The oxygen and CO exposures presented were not corrected for the enhanced exposures resulting from directional dosing. The U7A endstation and the procedures for temperatureprogrammed fluorescence yield near-edge spectroscopy (TPFYNES) experiments have been described in detail elsewhere.49 The U7A endstation is a two-chamber system with a reaction chamber capable of pressures up to 10 Torr on top, and a lower UHV chamber. TP-FYNES experiments were performed in the top chamber. The two chambers can be isolated by a gate valve and were pumped independently. The lower chamber is equipped with standard surface science instruments including a LEED/Auger system, QMS, and XPS. The Pt(411) sample was spot welded to two 0.5 mm Ta wires and mounted on a reentrant liquid nitrogen cooled feed trough that can be moved between the upper and lower chamber. The manipulator is capable of x-y-z directional and rotational motion. Temperature was measured with a chromel-alumel type K thermocouple spot welded to the back face of the sample. The crystal could be cooled to 100 K and heated resistively to 1000 K. Surface impurities were initially removed from the Pt(411)

CO Oxidation on a Stepped Platinum Surface

J. Phys. Chem. B, Vol. 109, No. 46, 2005 21849

Figure 2. Temperature-programmed desorption spectra of CO after various exposures at 100 K on Pt(411). Heating rate was 5 K/s. (a) CO TPD as a function of coverage: (a) 0.01 L; (b) 0.03 L; (c) 0.05 L; (d) 0.07 L; (e) 1.0 L; and (f) 5.0 L. At saturation (5 L), the ratio of terrace CO (410 K):step CO (520 K) is 40:60. (b) Partial desorption spectra of CO on Pt(411). The partial monolayer coverage was obtained by adsorbing the saturated 5 L coverage at 100 K, heated to 440 K to desorb terrace CO and cooled to 100 K before TPD.

surface by repeated cycles of argon ion sputtering and oxygen cleaning followed by annealing to 1000 K. The sample was cleaned in 10-3 Torr of oxygen followed by annealing to 1000 K between experiments. Surface cleanliness was verified by Auger spectroscopy and by monitoring the surface carbon in fluorescence yield experiments. The fluorescence yield near-edge spectroscopy (FYNES) soft X-ray spectra were measured using a proportional counter optimized for fluorescence detection of carbon.50 The surface carbon concentration was measured in the carbon continuum (330 eV) to determine the absolute coverage of carboncontaining species. The incident photon beam intensity was monitored using a gold mesh located directly upstream from the sample. All reported fluorescence signals measured by the FY detector were normalized to the incident photon beam intensity. For TP-FYNES experiments, the sample was background dosed in 2 × 10-7 Torr of CO at 100 K while measuring the carbon fluorescence yield to monitor CO surface coverage. When the surface carbon concentration reached saturation, the CO leak valve was closed and oxygen was backfilled at the desired pressure with a throttled turbomolecular pump. For oxygen pressures above 1 × 10-4 Torr, the ion gauge controller was turned off and a 1 Torr capacitance manometer was used to measure pressure. Once the oxygen pressure was stabilized, the sample temperature was increased linearly at 0.5 K/s while monitoring the carbon fluorescence yield. The fluorescence counts, incident current, and sample temperature were measured simultaneously during the experiments at 4-second intervals, resulting in a temperature resolution of (2.0 K. These experiments were performed with 450 µm/450 µm slits resulting in 1.2 eV energy resolution. Isothermal kinetic FYNES experiments were performed to develop a more complete picture of the CO + O2/Pt(411) reaction system. For these experiments, CO was adsorbed on the Pt(411) surface at 100 K. Oxygen was then introduced at the indicated pressure. Upon reaching the desired oxygen pressure, the sample was heated rapidly to the temperature for the isothermal experiment. This temperature was maintained

while monitoring the decay in the surface carbon coverage as a function of time. After oxidation was complete at the temperature being studied, the sample was heated to 600 K to oxidize the remaining adsorbed species. 3. Results 3.1. CO Desorption from the Pt(411) Surface. Temperatureprogrammed desorption spectra following the adsorption of CO on the clean Pt(411) surface at 100 K are shown in Figure 2a. At low exposure (0.01 L), CO desorbs in a single peak centered at 520 K. As the exposure of CO is increased to 0.07 L, an additional desorption feature develops at 410 K. Increasing the CO exposure further, the 410 and 520 K CO desorption peaks increase together until saturation coverage is reached (5 L). The presence of multiple desorption features for CO desorption from stepped surfaces has been previously reported.9,12,36 Based on previous work the high-temperature peak (520 K) is attributed to CO desorption from step sites and the low-temperature peak (410 K) to desorption from terraces. Using the Redhead method for first-order desorption and a preexponential factor of 1013 s-1,51 the desorption activation energy for the peak at 410 K is 25.0 and 32.0 kcal/mol for the peak at 520 K. Based on TPD peak deconvolution for a saturated monolayer the ratio of step to terrace CO is 60:40 on the Pt(411) surface. Desorption spectra from a partial CO monolayer formed by annealing a saturated monolayer to 440 K are shown in Figure 2b. Heating the CO-saturated surface to 440 K removes the low temperature 410 K CO desorption feature leaving more tightly bound CO, which desorbs at 520 K. The CO desorbing at 520 K has the same peak shape after partial desorption as the saturated CO layer desorbing from the Pt(411) surface. As discussed above, the CO peak at 520 K is associated with CO adsorbed on step sites.9,12,36 3.2. CO Oxidation on Pt(411). In the following section, data regarding the oxidation of CO will be presented. TPRS experiments for coadsorbed CO and oxygen will be discussed first. Then the reaction of preadsorbed CO in flowing oxygen pressures will be characterized as a function of oxygen pressure over the 10-9 to 10-7 Torr pressure range using mass spectro-

21850 J. Phys. Chem. B, Vol. 109, No. 46, 2005

Lewis et al.

Figure 3. TPRS of coadsorbed CO and oxygen on Pt(411). Oxygen and CO were both adsorbed at 100 K. Heating rate was 5 K/s. (a) Reaction spectra of preadsorbed CO coadsorbed with 5 L of O2 postexposure. One CO2 desorption feature at 325 K is observed only at low CO initial coverage (0.01 L). As the initial coverage of CO is increased oxygen adsorption is inhibited. (b) TPRS of preadsorbed 5 L of O2 with various postexposures of CO. For these oxygen-rich surface conditions, low-temperature CO2 formation is enhanced and four reaction channels at 120, 180, 280, and 380 K are observed.

metric detection of desorbing species. These low-pressure studies in flowing oxygen were extended to oxygen pressures of 0.002 Torr using FYNES to characterize the adsorbed CO coverages. Both TP-FYNES and isothermal kinetic experiments were performed to characterize the reaction kinetics in substantial detail. 3.2.1. Reaction of Coadsorbed CO + O and Peak Assignments. Figure 3 shows temperature-programmed reaction results over a range of initial oxygen and CO coverages. A complete TPRS study of CO oxidation on Pt(411) is being published separately.52 The primary results and peak assignments related directly to this study are summarized here for comparison. To characterize the effect of preadsorbed CO on dissociative oxygen adsorption, TPR spectra were obtained for several CO preexposures which were then postexposed to 5 L of oxygen (Figure 3a). Both the CO and O2 were adsorbed at 100 K. The 5 L oxygen exposure saturated the clean Pt(411) surface with oxygen (upper trace O2 panel) and no CO or CO2 was observed as expected. For the smallest 0.01 L CO preexposure, CO2 formation is observed in a broad peak centered at 325 K, while unreacted CO desorbs from step sites at 520 K and unreacted molecular oxygen desorbs at 145 K. For a 0.07 L CO preexposure, the surface is not quite saturated with CO, based on comparison with the CO saturated surface. However, the adsorption of oxygen under these conditions is almost completely inhibited as indicated by the small CO2 and O2 peaks marked with triangles in Figure 3a. For CO preexposures greater than 0.07 L, the saturated CO monolayer completely inhibits adsorption of oxygen and no CO2 is observed. TPRS results for a presaturated oxygen overlayer, postexposed to varying amounts of CO, are presented in Figure 3b. Increased amounts of CO2 are formed at low temperature with preadsorbed oxygen. Four major CO2 reaction peaks at approximately 120, 180, 280, and 380 K are observed depending on the CO postexposure. A 0.01 L CO postdose resulted in a

primary CO2 peak at ∼300 K with CO2 formation beginning near 150 K. With increasing CO postexposures, the CO2 yield increases and the temperature of the primary CO2 formation peak decreases (∼280 K). Postdosing 0.07 L of CO, the dominant peak temperature decreases to 280 K and additional CO2 peaks are formed at 120, 180, and 380 K. Increasing the CO postexposure to 5 L increases the CO2 yield below 200 K. For the 5 L CO postexposure, the 120 and 180 K peaks increase and a slight increase in the 280 and 380 K peaks is observed. Thus, when adsorbed oxygen is available for reaction CO2 formation begins at significantly lower temperatures especially for large CO postexposures. The TPRS peaks observed in Figure 3 have been assigned by comparison with published data and a complete TPRS study of CO oxidation on Pt(411).52 Based on previous work summarized in the Introduction and results presented for CO/Pt(411) TPD (see Figure 2), the 520 K CO peak is associated with steps and the CO peak at 410 K is associated with desorption from terrace sites.9,12,36 Based on Figure 2a, a CO exposure of 0.01 L results in CO desorption only from step sites. This step CO reacts with postadsorbed oxygen to form CO2 from 260 to 400 K. Low CO coverages reacting with oxygen to form CO2 at step sites has been observed on other stepped surfaces.30,33 With preadsorbed CO occupying both step and terrace sites, dissociative oxygen adsorption is inhibited limiting reaction yield significantly. In the case where oxygen is preadsorbed, large coverages of both CO and oxygen can be adsorbed and react as clearly shown by the large amount of CO2 formed in the upper trace in Figure 3b. These high coadsorbed coverages involve adsorption of CO and oxygen on both step and terrace sites. For CO postexposures of 0.07 L and above, CO2 formation below 200 K is significantly enhanced, while formation above 200 K increases only slightly. The low temperature oxidation peaks which appear below 200 K must therefore be associated with reaction at terrace sites.

CO Oxidation on a Stepped Platinum Surface

J. Phys. Chem. B, Vol. 109, No. 46, 2005 21851

Figure 4. Mass spectrometer based TPRS oxidation of step CO in flowing oxygen. Step CO was isolated by heating a saturated CO monolayer to 440 K. CO desorption from step sites is shown by the dotted line for reference. With CO adsorbed only on the step sites, CO2 is formed in multiple peaks over the 200-550 K temperature range.

Figure 5. Mass spectrometer based TPRS oxidation of saturated CO monolayer (step and terrace CO) in flowing oxygen. CO was adsorbed at 100 K before oxygen was introduced at the indicated pressure. CO desorption from clean Pt(411) is shown by the dotted line for reference. CO desorbs completely from terrace sites before reaction of step CO in low oxygen pressures.

The CO oxidation peaks above 200 K are associated with reaction primarily at step sites. 3.2.2. TPRS of Adsorbed CO in Flowing Oxygen Pressures. Mass spectrometer based TPR spectra of preadsorbed CO in flowing oxygen pressures are shown in Figures 4 and 5. For these experiments, CO was preadsorbed at 100 K. Oxygen was then introduced into the vacuum chamber until the required oxygen pressure was stabilized. The sample was then heated at 5 K/s while monitoring CO2 (m/e ) 44) and CO (m/e ) 28) desorption. The mass spectrometer ionizer was shrouded to minimize contributions from oxygen dissociated by the mass spectrometer filament. Temperature-programmed spectra for the reaction between a partial monolayer of preadsorbed CO in flowing oxygen pressures are shown in Figure 4. To prepare the partial CO monolayer, a saturated CO monolayer was heated to 440 K to desorb terrace CO. Thermal desorption spectra illustrating CO desorption from step sites (520 K) are shown in the right panel of Figure 4 (upper trace, right side). Note that the dominant CO desorption peak appears at 520 K for all these experiments indicating no terrace CO can be detected by TPRS experiments. The amount of CO2 formed increases with increasing oxygen pressure, as shown in the left panel of Figure 4. In 1 × 10-8 Torr of oxygen, the dominant CO2 peak appears at 530 K, with

significant peaks at 455, 310, and 257 K. In 5 × 10-8 Torr of oxygen, the dominant reaction peaks appear at 388 and 455 K with significant peaks at 512, 297, and 257 K. For oxygen pressures of 5 × 10-8 Torr and greater, the m/e ) 28 spectrum includes small peaks below 410 K which match the m/e ) 44 spectrum and are attributed to CO2 fragmentation. Increasing the oxygen pressure to 1 × 10-7 Torr of oxygen, the yield of CO2 from the surface increases with the dominant reaction peak at 410 K and significant peaks at 512, 378, 298, and 257 K. With increasing oxygen pressures, the dominant reaction channels move to lower temperature and increase in size. Taken together these results for a partial CO monolayer indicate the complexity of the reaction pathways available for oxidation of step CO. The next set of results reveals that in flowing oxygen the oxidation of a CO-saturated surface shows increased inhibition of dissociative oxygen adsorption. Figure 5 shows the oxidation of a saturated CO layer in oxygen pressures ranging from 1 × 10-8 to 1 × 10-7 Torr. For reference, CO desorption from the saturated Pt(411) surface is displayed as the dotted line. Saturated CO desorbs from the Pt(411) surface at 410 (terrace) and 520 K (step). In 1 × 10-8 Torr of flowing oxygen, the CO2 peak appears at 529 K. A substantial decrease in the high-temperature CO desorption peak at 520 K occurs with increasing oxidation. Increasing the oxygen

21852 J. Phys. Chem. B, Vol. 109, No. 46, 2005

Figure 6. TP-FYNES of step CO coverage in oxygen pressures ranging from 1 × 10-7 to 0.002 Torr. Spectra were taken in the carbon continuum at 330 eV.

Figure 7. TP-FYNES of saturated CO oxygen pressures ranging from 1 × 10-7 to 0.002 Torr. TP-FYNES of CO desorption is shown with the dashed line. Spectra were taken in the carbon continuum at 330 eV.

pressure to 5 × 10-8 Torr, both the high- and low-temperature CO peaks decrease and two CO2 reaction peaks appear at 505 and 535 K. Increasing the oxygen pressure to 1 × 10-7 Torr causes the yield of CO2 at 492 and 520 K to increase. On this CO-saturated surface, the oxidation peaks all appear above 410 K, the temperature where CO desorbs from terraces. With increasing oxygen pressure the yield of CO2 increases. Taken together these results indicate that preadsorbed CO inhibits oxidation for temperatures below the 410 K CO desorption temperature. Elevated oxygen pressures increase oxidation yields above 410 K and decrease the CO2 formation temperature. 3.3. TP-FYNES of Adsorbed CO in Flowing Oxygen Pressures. To extend the CO oxidation studies to higher oxygen pressures, the TP-FYNES experiments presented in Figures 6 and 7 were performed. For all results presented, fluorescence yield monitored in the carbon continuum (330 eV) was utilized to measure the coverage of CO as a function of sample temperature. The oxygen pressures for the TP-FYNES experi-

Lewis et al. ments represent the actual flux since background oxygen pressures were used. Thus, care should be exercised when comparing to the TPRS experiments since directional dosing was used for the TPRS experiments (see Experimental Section for details). TP-FYNES spectra for oxidation of partial CO monolayers of step CO in oxygen pressures from 10-7 to 10-3 Torr are shown in Figure 6. Step CO partial monolayers were prepared by partial desorption at 440 K as described previously. The desorption of step CO from the clean Pt(411) surface in a vacuum confirms that CO desorption occurs at 480 K for this heating rate (dotted trace). When heating the CO partial monolayer in oxygen, the onset of CO oxidation takes place well below the CO desorption temperature over the entire pressure range studied. In 1 × 10-7 Torr flowing oxygen, there is a sharp drop in surface CO coverage at 340 K, significantly lower than the 460 K initiation temperature for step CO desorption from the clean Pt(411) surface. The onset temperature for the oxidative removal of step CO continues to decrease with increasing oxygen pressure, and in 0.002 Torr flowing oxygen the oxidation temperature is reduced to 240 K. These results clearly indicate the important role played by oxygen pressure on this surface that contains open sites for oxygen adsorption. The next paragraph reports results for oxidation on the CO saturated surface where sites for oxygen adsorption are inhibited by adsorbed CO. TP-FYNES spectra for a saturated CO monolayer on the Pt(411) surface in oxygen pressures ranging from 10-7 to 10-3 Torr are shown in Figure 7. The desorption of CO from clean Pt(411) in a vacuum measured with TP-FYNES shows the same distinct low- and high-temperature CO desorption features attributed to CO desorption from step and terrace sites displayed in Figure 2 (top dotted trace in Figure 7). The decrease in CO desorption temperature in this experiment relative to the TPD experiments is the result of the decreased heating rate (0.5 K/s). The onset temperature for CO desorption from terrace sites is 360 K, and 460 K for step sites as shown in Figure 7. Heating in 10-7 Torr of oxygen pressure, the carbon monoxide coverage remains constant up to 360 K when a rapid decrease in CO surface coverage begins. This initial drop in carbon coverage is similar to the CO desorption spectrum for CO desorption in a vacuum (dotted line) and can be correlated with desorption of terrace CO. The CO coverage continues to follow the CO desorption spectrum up to 440 K where there is an abrupt drop in the CO surface coverage. This rapid drop in CO coverage at 440 K is associated with oxidation. Step CO is the dominant species on the surface at this temperature so that oxidation at 440 K is associated with step CO. Increasing the oxygen pressure to 10-6 Torr, the initiation temperature for oxidation decreases to 417 K. Again at this pressure, the desorption of terrace CO at 360 K precedes oxidation of step CO. For oxygen pressures above 10-6 Torr, the initiation temperature for CO oxidation continues to decrease below the 360 K terrace CO desorption temperature. In 10-4 Torr of oxygen, there is a single sharp drop in CO surface coverage at 350 K due to oxidation. In 0.002 Torr of O2 the initiation temperature for CO removal decreases further to 275 K and all CO is removed from the surface by 380 K. These studies with a CO-saturated surface highlight the multiple pathways and complexity of this “simple” oxidation reaction. 3.4. Isothermal FYNES of CO Oxidation in Oxygen Pressures. The reaction between CO adsorbed on the Pt(411) surface and flowing oxygen was further characterized by isothermal temperature jump experiments to determine kinetic

CO Oxidation on a Stepped Platinum Surface

Figure 8. Isothermal FYNES of saturated CO in 0.002 Torr of oxygen. The reaction rate increases with increasing temperature and decreasing CO surface coverage. Kinetic parameters over a range of CO coverages were determined using Arrhenius parameters as shown in Figure 9.

parameters (Figures 8-11). As shown in Figures 8-10 a series of isothermal studies were performed over the 10-6 to 10-3 Torr oxygen pressure range for oxidation of a presaturated CO monolayer (Figures 8 and 9) and for oxidation of CO initially adsorbed only at step sites (Figure 10). A series of oxygen reaction order experiments were performed over the 2 × 10-4 to 2 × 10-2 Torr oxygen pressure range for the initially COsaturated surface (Figure 11). Using these isothermal experiments, the oxidation activation energies, preexponential factors, and reaction order have been determined over an extended CO coverage range. Results from these kinetic studies are summarized in Table 1. Isothermal oxidation results for an initially saturated CO monolayer on Pt(411) in 0.002 Torr of oxygen are shown in Figure 8. As detailed in the Experimental Section, CO was first adsorbed at 100 K, the oxygen pressure was increased to 0.002 Torr, and the temperature was rapidly increased to the reaction temperatures of 305, 315, and 325 K. This isothermal method provides good signal to noise so that rates can be determined over an extended CO coverage. The initial rates above 0.85 ML have positive curvature and display clearly nonexponential behavior. As CO surface coverage decreases below 0.85 ML, the reaction rate accelerates. For coverages below 0.6 ML CO (only step CO on the surface), the isothermal kinetic data are characterized by first-order exponential decay. Rate constants were determined by measuring the slope over linear regions of semilog plots of CO coverage versus time over the coverage range indicated in Table 1. Arrhenius plots generated from isothermal spectra for a COsaturated surface oxidized in both 0.002 (Figure 8b) and 1 × 10-6 Torr (isothermals not shown) are shown in Figure 9. An activation energy of 24.1 kcal/mol and a preexponential factor of 5.8 × 1011 s-1 are obtained for the initially CO-saturated surface reacting over the 0.95-0.90 ML coverage range in 1 × 10-6 Torr of oxygen (Figure 9a). For an oxygen pressure of 0.002 Torr, the Arrhenius plot for the initially saturated surface reacting over the 0.95-0.90 ML coverage range results in an activation energy of 13.5 kcal/mol and a preexponential factor of 2.1 × 106 s-1 (Figure 9b). Over the 0.40 - 0.25 ML CO coverage range, the activation energy and the preexponential factor for the initially saturated CO monolayer reacting in 0.002

J. Phys. Chem. B, Vol. 109, No. 46, 2005 21853 Torr of oxygen are 7.8 kcal/mol and 2.1 × 103 s-1, respectively (Figure 9c). These results clearly indicate that the Arrhenius parameters for oxidation depend on both CO coverage and oxygen pressure for the initially CO-saturated surface. The kinetic parameters for oxidation of CO adsorbed only on step sites were determined in the same manner, and the isothermal kinetic data are shown in Figure 10a. Note again, the signal to noise even after reaction decreases the CO coverage. However, note in this case that the kinetic results appear to be consistent with a single-exponential decay process. For an oxygen pressure of 0.002 Torr, the Arrhenius plot for isothermal oxidation of step CO reacting over the 0.40-0.25 ML coverage range results in an activation energy of 10.6 kcal/ mol and a preexponential factor of 2.3 × 106 s-1 (Figure 10b). To establish the reaction order in oxygen, the oxidation of an initially CO-saturated surface was studied over the 2 × 10-4 to 2 × 10-2 Torr pressure range at 320 K (Figure 11). Note the dramatic enhancement in reaction rate as oxygen pressure is increased. The order plot over the 0.7 to 0.5 CO coverage range indicates that the reaction order is 0.63. The order plot taken over the 0.5 to 0.3 CO coverage range indicates that the reaction order is 0.75. Taken together these results indicate that the reaction order for oxidation of CO on this stepped platinum surface depends on CO coverage. 4. Discussion The oxidation of adsorbed CO on the stepped Pt(411) surface has been experimentally characterized in flowing oxygen pressures ranging from 10-8 to 0.002 Torr. The combination of TP-FYNES and isothermal kinetic experiments with TPRS provide interesting new insight into the behavior of this important reaction system on a stepped surface. For oxygen pressures above 10-5 Torr, low-temperature CO oxidation is observed, even for a saturated monolayer of CO in contrast to results observed on low-index surfaces. This new result indicates that inhibition of oxygen adsorption by CO is not dominant for elevated oxygen pressures. 4.1. Inhibition of CO Oxidation by Large Coverages of CO and Low Oxygen Pressures. For the CO-saturated surface reacting in oxygen pressures below 10-5 Torr, two narrow CO2 formation peaks are observed near 500 K over a 25 K temperature range (Figure 5). Both CO2 peaks appear above the 410 K desorption temperature for terrace CO. Thus terrace CO desorption, which frees sites for dissociative oxygen adsorption at vacant sites appears to be the rate-limiting step for high CO coverages and low oxygen pressures. The narrow temperature separation (25 K) for CO2 formation and large yields suggest similar energetics for the rate-limiting processes for oxidation of both step and terrace CO. The preferential depletion of carbon monoxide adsorbed at step sites (compare spectra a and b in Figure 5) to form CO2 at 530 K appears to support the preferential reaction of step CO. However, the surface mobility of CO at these temperatures coupled with preferential occupation of the strongest sites appears to be the cause of the correlation. To summarize, the results in Figure 5 indicate that for large coverages of preadsorbed CO reacting with a low pressure of oxygen the reaction is limited by the desorption of CO and the enhanced oxygen adsorption associated with these vacant sites. This observation is consistent with many previously published low-pressure studies of CO oxidation for the CO-saturated surface and low oxygen pressures. Dissociative oxygen adsorption is not inhibited when terrace CO is desorbed from the surface prior to reaction (Figure 4). In these cases where only step CO is initially present, oxidation

21854 J. Phys. Chem. B, Vol. 109, No. 46, 2005

Lewis et al.

Figure 9. Arrhenius parameters for isothermal FYNES of saturated CO in oxygen pressures. (a) Saturated CO in 1 × 10-6 Torr of oxygen. CO coverage range ) 0.95-0.90 ML. Ea ) 24.1 kcal/mol, A ) 5.8 × 1011 s-1. (b) Saturated CO in 0.002 Torr of oxygen. CO coverage range ) 0.95-0.90 ML. Ea ) 13.5 kcal/mol, A ) 2.1 × 106 s-1. (c) Saturated CO adsorbed in 0.002 Torr of oxygen. CO coverage range ) 0.40-0.25 ML. Ea ) 7.8 kcal/mol, A ) 2.1 × 103 s-1.

Figure 11. Isothermal FYNES of saturated CO at 320 K in varying oxygen pressures. Saturated CO was adsorbed at 100 K and oxygen was introduced at the pressure indicated.

Figure 10. Isothermal FYNES of step CO in 0.002 Torr of oxygen. (a) Isothermal FYNES spectra monitoring CO coverage as a function of time at constant temperature (dotted line). The FYNES spectra are fit well with first-order exponential curves (solid line). (b) Kinetic parameters determined using the Arrhenius method. CO coverage range ) 0.40-0.25 ML. Ea ) 10.6 kcal/mol, A ) 2.3 × 106 s-1.

proceeds for temperatures as low as 200 K. For these partial coverage conditions, vacant sites for dissociative oxygen

adsorption are available and multiple CO2 oxidation pathways in the 200 to 550 K temperature range appear. Compared to the CO-saturated surface (Figure 5), CO2 is formed at considerably lower temperature since terrace sites are available for dissociative oxygen adsorption. The formation of CO2 near 200 K is also enhanced when preadsorbed oxygen is exposed to CO resulting in large coadsorbed coverages of CO and oxygen (Figure 3). Typically, CO at terraces is more reactive than CO adsorbed at step sites.22,25,36 At elevated temperatures, CO diffusion from the steps to terrace sites has been previously proposed to rationalize low-temperature oxidation of step CO and oxygen.12 Step CO was displaced to terrace sites for a partial monolayer of CO at steps and terraces on Pt(335).30 In these experiments it is quite clear that vacant terrace sites enhance low-temperature CO oxidation. On the basis of previous studies, increased rates of dissociative oxygen adsorption play an important role in the enhancement of CO oxidation. 4.2. Enhanced Low-Temperature CO Oxidation with Increasing Oxygen Pressure. The initiation temperature for CO oxidation decreases with increasing oxygen pressures for

CO Oxidation on a Stepped Platinum Surface

J. Phys. Chem. B, Vol. 109, No. 46, 2005 21855

TABLE 1: Summary of the Arrhenius Parameters Determined in Figures 8-10a surface Pt(411)

Pt/Al2O3 Pt(111)

initial CO coverage (ML)

O2 reaction pressure (Torr)

1.0 1.0 1.0 0.6 1.0 1.0 1.0

1 × 10-6 2 × 10-3 2 × 10-3 2 × 10-3 2 × 10-3 2 × 10-3 2 × 10-3

CO coverage range (ML) 0.95-0.9 0.95-0.9 0.4-0.25 0.4-0.25 >0.7